Downloads · 30 days
3
10% of all-time downloads
Aarushhh_-_untrained-pruned-customffn-llama3.1-8b-4bits
RichardErkhov/Aarushhh_-_untrained-pruned-customffn-llama3.1-8b-4bits
Aarushhh_-_untrained-pruned-customffn-llama3.1-8b-4bits is a machine learning model from RichardErkhov. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
All-time downloads
29
Public
Parameters
2.7B
3 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors3 GB · 99%
Parameter types
How the weights are stored.
U81.6B · 59%
Model card
From the Hugging Face model README
Quantization made by Richard Erkhov.
untrained-pruned-customffn-llama3.1-8b - bnb 4bits
- Model creator: https://huggingface.co/Aarushhh/
- Original model: https://huggingface.co/Aarushhh/untrained-pruned-customffn-llama3.1-8b/
Original model description:
base_model:
- meta-llama/Llama-3.1-8B library_name: transformers license: cc-by-nc-sa-4.0
An (untrained) pruned version of llama 3.1 8b with a bigger intermediate size
<!-- Provide a quick summary of what the model is/does. -->Model Details
The model has 4 layers which are the 1st, 2nd, 3rd and the last layers of llama 3.1 8b
The embeddings and the lm_head have been also been copied over
The intermediate size is 28672 (same as llama 3.1 70b) vs the 14336 of llama 3.1 8b
The model still needs to be trained
This model should be used for research purposes only
Model Card Author
Aarush Khilosia
- Type
- llama
- Created
- Mar 8, 2025
- Updated
- Mar 8, 2025